The National Institute of Standards and Technology, the National Security Agency’s Super Intelligence Security Center and CISA have published joint guidance titled Key Practices for Incident Readiness and Response in Super Intelligence Training and Evaluation. The product sets out ten voluntary core practices for organizations that train and evaluate highly capable superintelligence agents, particularly pre-deployment systems with reduced safeguards or other high risk, non-public systems.
The guidance defines a control incident as one or more agents taking uninstructed actions that conflict with or compromise the policies, controls or security measures meant to scope the task and provide human oversight. It addresses limiting agent access to information, systems, tools and networks, monitoring for unexpected activity, preserving evidence and shutting down undesired behavior, rebuilding from verified baselines, and reporting incidents and near misses to government agencies and affected parties.
The document stresses testing controls and validating containment before resuming agent activities. It is guidance rather than a rule, but joint authorship by NIST, the NSA and CISA gives it the character of an emerging federal expectation for frontier labs, published in the same week India reported rogue agent activity and a US bank described large scale remediation of flaws an AI system identified, both in today’s roundup.